1. Field of the Invention
The present invention relates to an imaging device and a dynamic range enhancement method and apparatus therefore to improve image quality.
2. Description of the Related Art
An image sensor is a device that converts an optical image to an electric signal. Typically, an image sensor comprises a Charge-Coupled Device (CCD) or a Complementary Metal-Oxide-Silicon (CMOS) sensor. A CCD image sensor typically serializes parallel analog signals, which are in arrays of photoelectric light sensors by capacitors transferring an electric charge of each pixel to a common output structure which converts the charge to a voltage and buffers and sends the voltage off-chip. On the other hand, in a CMOS image sensor, pixels represented by respective MOS transistors convert the charges to voltages and the voltages are output in accordance with switching operations of the transistors.
One of the most important factors in determining product quality of an imaging device is its dynamic range of the pixels. Dynamic range is a ratio of a pixel's saturation level to its signal threshold. Typically, the dynamic range refers to the maximum signal power that the device can tolerate without distortion of the input signal. In an image sensor, as the dynamic range of a device is increased, the ability to measure the dimmest intensities in an image is improved, resulting in good image quality.
In order to expand the dynamic range, a signal variation-based dynamic range expansion method has been used with limited success. The dynamic range expansion (DRE) method differentiates a signal and adjusts the differentiated signal to expand the dynamic range. In the DRE method, the dynamic range characteristic of the signal changes by varying the differential value. This signal variation-based dynamic range expansion method is useful for one-dimensional voice signals. However, this method is not suited for processing two-dimensional images signal due to large calculations of quadratic calculus and associated iterations.
Another conventional dynamic range expansion method uses multiple images taken at different exposure levels. This particular method estimates a response function of the image sensor using multiple images taken at different exposure levels and compensates the image by taking an inverse function of the response function. However, this method has several drawbacks, such as requiring a large memory for storing the multiple images, and the iterative process and is restricted for applying to real time applications.